Can you validate my understanding on the architecture for Primary Server fault tolerance and offer any advice if there are improvements I can make? We have 2 primary servers with an external SQL DB server. We have the agent in the base of our PC and VDI deployments (unregistered) and register the agent via command line to the first primary server. In testing with logged in workstations and active zen icons, I am shutting down the first primary server and I see that the icons are still active (assuming this is due to cached settings). As expected, any new imaged workstations or recomposed VDI VMs, it fails due to not being able to communicate to the first primary server and register.

In a fault tolerant scenario like this, is my only option then to go with MLB or an L4 switch in order to fail over to the second Primary server seamlessly? I need a solution that would still allow registering to the second primary server in case the first is down. Thanks for confirming and your input.